Effects of native language and speech rate on perceptual and decisional processing of voicing and syllable affiliation in stops

Abstract

Previous work shows that variation in speech rate influences the perception of voicing distinctions (/b/-/p/) and syllable affiliation (“pea”-“eep”), and it is well-documented that native language influences how listeners perceive phonological distinctions. We analyze the influences of speech rate and native language in the perception of voicing and syllable affiliation by applying a model of perception and response selection to data from Japanese, English, and Korean listeners who identified the voicing and the syllable affiliation of (English) stops produced at slow, moderate, and fast rates. The fitted model indicates that for all three native language groups, perceptual salience decreases substantially as speech rate increases for both voicing and syllable affiliation. Even at slow rates, however, the salience of voicing is lower for coda than for onset stops. In addition, as rate increases, all three groups exhibit an increasing bias toward “onset” responses, with a bias toward “voiced” responses for coda stimuli and toward “voiceless” responses for onset stimuli. Despite broad similarities across all three groups, fine-grained patterns of perceptual salience and response bias vary with listeners’ native language. These data and fitted models illustrate the utility of rate-varied speech in investigations of native language effects in speech perception.